| { | |
| "model_type": "turing_rdlm", | |
| "architectures": [ | |
| "TuringLM" | |
| ], | |
| "vocab_size": 32000, | |
| "D_MODEL": 6144, | |
| "D_U": 3072, | |
| "D_V": 3072, | |
| "HIDDEN_REACT": 6144, | |
| "N_ATTRACTORS": 24576, | |
| "MAX_LEN": 128, | |
| "pad_token_id": 0, | |
| "custom_architecture": true | |
| } |
| { | |
| "model_type": "turing_rdlm", | |
| "architectures": [ | |
| "TuringLM" | |
| ], | |
| "vocab_size": 32000, | |
| "D_MODEL": 6144, | |
| "D_U": 3072, | |
| "D_V": 3072, | |
| "HIDDEN_REACT": 6144, | |
| "N_ATTRACTORS": 24576, | |
| "MAX_LEN": 128, | |
| "pad_token_id": 0, | |
| "custom_architecture": true | |
| } |